Aldose Reductase Deficiency Reduced Vascular Changes in Neonatal Mouse Retina in Oxygen-Induced Retinopathy

Abstract

Purpose.: Retinal neovascularization is the major pathologic process in many ocular diseases and is associated with oxidative stress. Deficiency of aldose reductase (AR), the first enzyme in the polyol pathway for glucose metabolism, has been shown to reduce oxidative stress and blood vessel leakage. The present study aimed to investigate the effect of AR deficiency on retinal neovascularization in a murine oxygen-induced retinopathy (OIR) model.

Methods.: Seven-day-old wild-type (WT) and AR-deficient (AR^−/−) mice were exposed to 75% oxygen for 5 days and then returned to room air. Vascular obliteration, neovascularization, and blood vessel leakage were analyzed and compared. Immunohistochemistry for AR, nitrotyrosine (NT), poly(ADP-ribose) (PAR), glial fibrillary acidic protein (GFAP), and Iba-1, as well as Western blots for vascular endothelial growth factor (VEGF), phospho-Erk (p-Erk), phospho-Akt (p-Akt), and phospho-IκB (p-IκB) were performed.

Results.: Compared with WT OIR retinae, AR^−/− OIR retinae displayed significantly smaller central retinal vaso-obliterated area, less neovascularization, and reduced blood vessel leakage. Significantly reduced oxidative stress and glial responses were also observed in AR^−/− OIR retinae. Moreover, reduced microglial response in the avascular area but increased microglial responses in the neovascular area were found with AR deficiency. Furthermore, expression levels of VEGF, p-Erk, p-Akt, and p-IκB were significantly reduced in AR^−/− OIR retinae.

Conclusions.: Our observations indicated that AR deficiency reduced retinal vascular changes in the mouse model of OIR, indicating that AR can be a potential therapeutic target in ischemia-induced retinopathy.

Introduction

Ischemic retinopathy is a leading cause of blindness worldwide. It is common in diabetes, retinal vein occlusion, and retinopathy of prematurity. ¹ The well-known features of this disease include retinal ischemia, increased vascular permeability, and neovascularization. ¹ Retinal neovascularization is characterized by the development of sprouts from retinal vessels. In most cases, these newly formed vessel sprouts can penetrate the inner limiting membrane (ILM) and grow into the vitreous, leading to tractional retinal detachment and eventually blindness. ²  

Several angiogenic growth factors and cytokines that stimulate angiogenesis are identified in the pathogenesis of neovascularization. ^3,4 Vascular endothelial growth factor (VEGF), an endothelial cell–specific mitogen and a vasopermeability factor, plays a key role in both normal and abnormal retinal vascular growth. ^5,6 Disruption of Müller cell–derived VEGF inhibited ischemia-induced neovascularization, vascular leakage, and blood–retinal barrier (BRB) breakdown. ⁷ In patients and experimental animals, the induction of VEGF expression by ischemia has been strongly associated with oxidative stress. ^8–11 In addition, the role of inflammatory cytokines including tumor necrosis factor (TNF)-α, interleukin (IL)-1, and IL-8 has been emphasized in inflammation-associated angiogenesis. ^12–15 Moreover, increased microglia and macrophages were shown to contribute to ischemia-induced retinal neovascularization and neovascular tuft regression in the mouse retina. ¹⁶ Previous studies also demonstrated a link between intracellular oxidative stress and retinal inflammatory diseases. ^17–19  

Diabetic retinopathy (DR) is one example of ischemic retinopathy. The hallmarks of this disease include basement membrane thickening, loss of pericytes, BRB breakdown, and neovascularization. ²⁰ Current treatments to inhibit the proliferative stage of DR include laser and vitrectomy, anti-VEGF therapies, and steroids but not without limitations and side effects. ^15,19 Therefore, there has been a continuing effort to understand the molecular mechanisms that contribute to the changes observed in the diabetic retina. Increased aldose reductase (AR) activity and oxidative stress as well as inflammation have been implicated in the pathogenesis of DR. ^15,19–21  

AR is the first enzyme in the polyol pathway for reducing glucose to sorbitol, using nicotinamide adenine dinucleotide phosphate (NADPH) as a cofactor. ²² Increased AR activity is thought to contribute to oxidative stress. ^23–28 Indeed, AR deficiency or inhibition of AR activity could reduce ischemia-induced oxidative stress in mouse retina. ²⁹ Genetic deletion of AR protected against the diabetes-induced BRB breakdown ²⁰ while pharmacological inhibition of AR activity with Fidarestat prevented retinal oxidative stress and VEGF overexpression. ³⁰ In addition, pharmacological inhibition or genetic ablation of AR could attenuate the inflammatory signals triggered by cytokines, growth factors, endotoxins, and high glucose in cell cultures and animal models of various inflammatory diseases. ^31,32 However, direct evidence to support the role of AR in neovascularization, a key pathological feature in ischemic retinopathy, is still lacking. 

The aim of our study was to investigate the effect of genetic deletion or pharmacological inhibition of AR on retinal neovascularization. Our hypothesis was that AR deficiency might reduce neovascularization as a consequence of reducing oxidative stress, thereby attenuating VEGF overexpression and modulating inflammation. We used the well-established animal model of oxygen-induced retinopathy (OIR) with hyperoxia-induced vaso-obliteration and the subsequent relative-hypoxia–triggered neovascularization to test our hypothesis. ⁴  

Methods

Animals

All the experimental and animal handling procedures were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The use of animals was conducted according to the requirements of the Cap. 340 Animals (Control of Experiments) Ordinance and Regulations, and all relevant legislation and Codes of Practice in Hong Kong, and was approved by the Faulty Committee on the Use of Live Animals in Teaching and Research in The University of Hong Kong (CULATR 1691–08 and 2423–11). 

Animal Model of OIR

Generation of AR-deficient (AR^−/−) mice was achieved using embryonic stem (ES) cell technology. ³³ AR^−/− mice were backcrossed to the C57BL/6N strain to the 11th generation (N11) and were considered to be congenic with C57BL/6N. AR^−/− mice showed growth rate and reproductive capacity similar to those of C57BL/6N mice. ³³  

OIR was induced in C57BL/6N wild-type (WT) mice and AR^−/− mice. Neonatal mice and their nursing dams were exposed to 75% oxygen (PRO-OX110 chamber controller; Biospherix Ltd., New York, NY) between postnatal day 7 (P7) and P12 and then returned to room air on P12. ⁴ During the exposure to high oxygen, soda lime was placed inside the chamber to serve as CO₂ quencher. ³⁴ In a single day, only one WT litter and one AR^−/− litter were exposed to high oxygen. To reduce the runty phenotype, the litter size was limited to seven or eight pups for each mother. ³⁴ To overcome biological variability, pups from at least four litters were used. In each litter, the pups were further divided into different parties for various analyses. For example, in each litter two or three pups were randomly selected for retinal flat mounts; two or three pups were used for retinal section preparation while two or three pups were used for protein extraction. For drug treatment, half of the pups from each litter received daily intraperitoneal injection of Fidarestat (2 mg/kg body weight ²⁹ dissolved in distilled H₂O [dH₂O]; Sanwa Kagaku Kenkyusho Co., Ltd., Nagoya, Japan), and the other half received dH₂O as control from P12 through P17. On P17, samples were collected, and only pups with weight ≥6 g were used. ³⁴ Both eyes were enucleated and fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS; 0.01 M, pH 7.4) at 4°C. One eye was processed for subsequent histological analysis. Retinal flat mount was obtained from the other eye for analysis of vascular changes. For Western blot analysis, two retinae from one mouse were pooled and kept at −80°C until use. 

Retinal Morphology

Vasculature Immunohistochemistry in Retinal Flat Mounts.

Flat-mounted retinae were fixed for 1 to 2 hours and permeabilized in ice-cold ethanol (70%) for 20 minutes followed by PBS with 1% Triton X-100 (Sigma-Aldrich Corporation, St. Louis, MO) for 30 minutes. The retinae were washed and incubated with Alexa Fluor-488-conjugated isolectin GS-IB4 from Griffonia simplicifolia (0.2 mg/ml; Invitrogen Corporation, Carlsbad, CA) overnight at 4°C. The retinae were rinsed three times in PBS and coverslipped. The superficial, middle, and deep layers of the retinal vasculature on P7, P12, and P17 in room-air controls were investigated using a confocal laser scanning microscope under 200× magnification (LSM700; Carl Zeiss, Thornwood, NY). The three distinct retinal vascular layers were observed and determined by adjustment of the microscope focus. 

Retinal Structure and Retinal Layer Thickness.

Eyeballs were enucleated and fixed with 4% PFA overnight at 4°C, dehydrated with a graded series of ethanol and chloroform, and finally embedded in paraffin. Cross sections (6 μm) were cut through the cornea parallel to the optic nerve using a microtome (HM 315R; Microm, Heidelberg, Germany). Deparaffinized retinal sections with optic nerve head were stained with hematoxylin and eosin (H&E) for the examination of retinal structure on P12, P14, and P17. Photomicrographs were captured with an upright microscope (Eclipse 80i; Nikon, Tokyo, Japan) under 200× magnification for retinal structure and 400× magnification for retinal layer thickness measurement at both central and midperipheral retinal areas. Central retinal area was defined as the area ∼100 μm away from the optic nerve; midperipheral retinal area was defined as the middle area between the optic nerve and the most peripheral part of the retina on each side. Thickness of the total layer (from ILM to outer edge of outer nuclear layer [ONL]), ONL, inner nuclear layer (INL), and inner plexiform layer (IPL) was then measured using Spot Advanced software (SPOT Imaging Solutions, Sterling Heights, MI). 

Central Vaso-Obliteration

Flat-mounted retinae were incubated with Alexa Fluor-488-conjugated isolectin GS-IB4 from Griffonia simplicifolia (0.2 mg/ml; Invitrogen) as mentioned above. Images of the superficial blood vessel layers were captured with an upright microscope (Eclipse 80i, Nikon) under 40× magnification. The area of central vascular obliteration was measured using Sigma Scan Pro software (SPSS, Chicago, IL). During the analysis, care was taken not to include areas in which a dissection or mounting artifact was present. The borders of the avascular areas were traced. The area of central obliteration and total retina in pixels were computed, and the percentage of central obliterated area over total retinal area was calculated. 

Neovascularization

Quantitative Assessment of Retinal Neovascularization by Fluorescein Angiography.

Based on the higher intensity of isolectin staining and the characteristic appearance of neovascular tufts, isolectin-stained flat-mounted retinae were used for measuring the tuft area as previously published. ³⁵ Briefly, the tuft area was first selected by hand with the Paint Bucket tool (Photoshop CS5; Adobe Systems, Seattle, WA) and then measured by adjusting the threshold in Image J (National Institutes of Health, Bethesda, MD). In addition, the total surface area of the retina was outlined using the outermost vessel of the arcade near the ora serrata as the border. The percentage of neovascular area over total retinal area was calculated and compared between WT and AR^−/− OIR groups. 

Quantitative Assessment of Retinal Neovascularization by Counting Vascular Lumens.

Retinal sections with optic nerve head were excluded to eliminate the error caused by normal vessels extending from the optic nerve. ⁴ A total of eight sagittal sections, each 30 μm apart, four on each side of the optic nerve, were stained with H&E to assess retinal vasculature with an upright microscope (Eclipse 80i, Nikon) under 400× magnification. Vascular lumen extending from the retina into the vitreous was counted. ³⁶ The average intravitreal vessels/section was calculated for each group. 

IgG Extravasation

Deparaffinized retinal sections with optic nerve head were subjected to antigen retrieval by incubation with proteinase K followed by the Mouse on Mouse biotinated anti-mouse IgG secondary antibody and avidin-biotin-peroxidase complex (Vector Laboratories, Inc., Burlingame, CA). IgG immunoreactivity was visualized by incubation with the substrate, diaminobenzidine (Zymed Laboratories, Inc., South San Francisco, CA). Subsequently, the sections were washed and counterstained with hematoxylin. IgG staining outside the blood vessel lumen indicated leakage as a result of BRB breakdown. ²⁰ The number of leaky vessels in the GCL and OPL was quantitated with an upright microscope (Eclipse 80i; Nikon) under 600× magnification. Three sections were selected for each sample. 

Immunohistochemistry

Deparaffinized retinal sections with optic nerve head were subjected to antigen retrieval by incubation with proteinase K. Sections were blocked with normal serum and incubated with primary antibodies: rabbit anti-aldose reductase (AR 1:400; Biorbyt, Cambridge, UK), rabbit anti-glial fibrillary acidic protein (GFAP 1:500; Dako, Glostrup, Denmark), rabbit anti-nitrotyrosine (NT 1:200; Upstate Biotechnology, Lake Placid, NY), and mouse anti-poly(ADP-ribose) (PAR 1:200; Alexis, Lausen, Switzerland), respectively, at 4°C overnight. GFAP staining was visualized by reacting with Alexa Fluor-568 goat anti-rabbit IgG (1:500; Molecular Probes, Invitrogen) for 1 hour at room temperature. After washing with PBS, the sections were coverslipped with fluorescent mounting medium (Dako, Carpinteria, CA). For visualization of AR, NT, and PAR immunoreactivity, sections were further incubated with biotinated goat anti-rabbit secondary antibody (1:200) and the avidin-biotin-peroxidase complex (Vector Laboratories, Inc.). The immunoreactive signal was developed using diaminobenzidine (Zymed Laboraotories, Inc., South San Francisco, CA). The sections were then washed and coverslipped for examination with an upright microscope (Eclipse 80i, Nikon) under 400× magnification for GFAP and NT and 600× magnification for AR and PAR. Semiquantitative analysis was used to assess the immunoreactivity as previously described. ^37,38 Briefly, all retinal sections for analysis were processed at the same time in a single round of the immunohistochemistry (IHC) experiment. After the immunohistochemical procedures, microscopic slides were randomly coded and examined in a blinded approach. IHC scores were given based on the intensity and location of the staining along the whole retina a score of 1 was assigned for the weakest immunoreactivity, and a score of 5 was assigned for the strongest immunoreactivity. Retinal sections were then decoded, and the scores were compared among the experimental groups. Photomicrographs were captured at both central and midperipheral retinal areas. 

Assessment of Retinal Microglia

Flat-mounted retinae were fixed for 2 hours in 4% PFA and permeabilized in ice-cold ethanol (70%) for 20 minutes followed by PBS with 1% Triton X-100 for 30 minutes. The retinae were washed and blocked with 10% goat serum for 1 hour at room temperature and incubated with rabbit anti-Iba-1 (1:800; Wako Chemicals USA, Richmond, VA) and Alexa Fluor-488-conjugated isolectin GS-IB4 from Griffonia simplicifolia (0.2 mg/ml; Invitrogen) for 3 days at 4°C. Signal of Iba-1 immunoreactivity was visualized by reacting with Alexa Fluor-568 goat anti-rabbit IgG (1:500; Molecular Probes; Invitrogen) for 1 hour at room temperature. Confocal images of the central and midperipheral zone from superficial and deep vascular layers of the retina were collected using a confocal laser scanning microscope (LSM700; Carl Zeiss, Thornwood, NY). The number of microglia cells in eight selected 200X fields of view was determined for each zone. 

Western Blot Analysis

Protein was extracted from P17 retinae using RIPA lysis buffer (0.15 M NaCl, 5 mM EDTA [pH 8.0], 1% Triton X-100, 10 mM Tris [pH 7.4]) supplemented with fresh protease inhibitor (Roche, Indianapolis, IN) and phosphatase inhibitor (Calbiochem, Darmstadt, Germany) (lysis buffer:protease inhibitor:phosphatase inhibitor in a ratio of 50:1:1). Two retinae from one animal were pooled in 60 μL supplemented RIPA lysis buffer, sonicated for 15 seconds and kept on ice for 30 minutes. The sample was then centrifuged at 4°C and 13,500 rpm for 30 minutes. The supernatant was collected as the total protein lysate, quantified by Bio-Rad DC Protein Assay (Bio-Rad, Hercules, CA), and stored at −80°C. Protein lysate (15 μg) together with a marker lane (Full Range Rainbow Molecular Weight Markers, Amersham Bioscience, Pittsburgh, PA) was separated by SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membrane (Millipore, Billerica, MA). The membrane was blocked with 5% milk in tris-buffered saline with 0.1% Tween 20 (TBST) for 1 hour. The membrane was incubated with mouse anti-VEGF antibody (1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), mouse anti-p-Akt (Ser473) (1:1000), rabbit anti-Akt (1:2000), mouse anti-p-Erk1/2 (1:1000), mouse anti-Erk1/2 (1:2000), mouse anti-p-IκB (1:1000), and rabbit anti-IκB (1:1000) antibodies, respectively, from Cell Signaling Technology (Danvers, MA) overnight at 4°C. The blot was then washed with TBST twice for 10 minutes and tris-buffered saline (TBS) once for 10 minutes, and probed with corresponding secondary antibody for 1 hour at room temperature. Protein bands were visualized using ECL Chemiluminescence Reagent (GE Healthcare, Buckinghamshire, UK) whose intensities were quantitated by Image J, normalized against beta-actin, and compared. 

Study Design and Statistical Analysis

All experiments and analyses were performed in a double-blinded manner. The immunohistochemical investigations for all retinal sections using the same antibody were performed in one single experiment to eliminate interexperiment errors and variation as described above. Unpaired t-test, one-way ANOVA, and Bonferroni's multiple comparison test, Kruskal-Wallis test, and Dunn's multiple comparison test were used for comparison of results as specified (Prism v5.0; GraphPad Software, Inc., San Diego, CA). Statistically significant difference was set at P < 0.05. Data are presented as mean ± SEM. 

Results

Retinal Morphology

AR^−/− mice have been previously shown to have growth rate and reproductive capacity similar to those of C57BL/6N mice. ³³ The development of retinal vasculature in WT and AR^−/− mice was investigated using endothelial cell marker isolectin GS-IB4 (Fig. 1). Generally, the superficial vascular plexus was first sparsely formed in the retinae of both genotypes. Subsequently, blood vessels branched from the superficial layer to the deep layer of the retina to form the deep vascular plexus in a dense manner during the second week after birth. Finally, an intermediate vascular plexus developed during the postnatal third week. In addition, comparable retinal structure was observed in H&E-stained WT and AR^−/− room-air (RA) retinal sections on P12, P14, and P17 (Fig. 2A). No difference could be observed in WT and AR^−/− RA retinae (Fig. 2A). Moreover, no significant difference was observed in the thickness of the total retinal layer, ONL, INL, and IPL between WT and AR^−/− RA retinae on P12, P14, and P17 (Fig. 2B). 

Figure 1. 

 

Isolectin GS-IB4-stained retinal vasculature in different layers of WT and AR^−/− RA retinae. Green fluorescence indicated isolectin-stained endothelial cells in blood vessels. The time course and formation of vascular plexus were similar in WT and AR^−/− RA retinae. On P7, blood vessels were observed only in the superficial layer of the retina. On P12, dense deep vascular plexus was formed and an intermediate layer appeared. On P17, a well-formed intermediate layer was observed. S, superficial; M, middle; D, deep. Arrows: vessel branching points indicated by green dots with higher fluorescence. n = 4 to 8 for each group. Scale bar: 500 μm in whole retina on P7; 100 μm in deep peripheral retina on P17.

Figure 2. 

 

H&E-stained WT and AR^−/− RA retinal sections and measurement of retinal layer thickness (A, B). Retinal structure was similar in WT and AR^−/− RA retinae on P12, P14, and P17 (A). No significant difference was observed in total retinal layer, ONL, INL, and IPL thickness between WT and AR^−/− RA retinae on P12, P14, and P17 (B). n = 6 for each group. One-way ANOVA followed by Bonferroni's multiple comparison test. Scale bar: 50 μm.

Expression of AR

Our current observations demonstrated that under normal conditions, AR expression was mainly located in the INL of mouse retina (arrowheads, Figs. 3A, 3C) and that there was faint immunostaining in the GCL (arrows, Figs. 3A, 3C) and cells surrounding the blood vessels (double arrows, Fig. 3E). After OIR, more intense AR immunoreactivity was observed in the GCL (arrows, Figs. 3B, 3D) and neovascular tufts (double arrows, Fig. 3F), while minimal/no change in AR immunoreactivity was found in the INL (arrowheads, Figs. 3B, 3D). 

Figure 3. 

 

Immunohistochemical staining using AR antibody in WT retinae on P17 (A–F). AR immunoreactivity was mainly located in cells in INL in RA controls (arrowheads, A, C) and was faint around blood vessels (double arrows, E). After OIR, AR immunoreactivity was observed in GCL (arrows, B, D) in addition to INL (arrowheads, B, D). Moreover, AR immunoreactivity was increased in the neovascular tufts in WT OIR retinae (double arrows, F). Scale bar, 25 μm (D, F).

Retinal Vaso-Obliteration

In our experiment, AR^−/− mice were observed to breed well, and the weight of pups on P7 (∼4 g) and P17 (∼6 g) was identical to that of WT mouse pups with limited litter size in OIR. After hyperoxia exposure, the central avascular zone indicating blood vessel obliteration was observed in both genotypes (Figs. 4B–E, 4G–J). After OIR, the central vaso-obliterated area was largest on P12 (Fig. 4M) and became gradually smaller on P14, P15, and P17. At all time points examined, the avascular area in AR^−/− OIR retinae was significantly smaller than that in WT OIR retinae (Fig. 4M). In addition, a faster drop in central avascular area from P12 to P15 in AR^−/− OIR retinae (Fig. 4M) indicated occurrence of an early revascularization. Moreover, significantly reduced central avascular area was also observed in Fidarestat-injected retinae versus dH₂O-injected controls (Figs. 4K, 4L, 4N). 

Figure 4. 

 

Flat-mounted retinae stained with isolectin GS-IB4 (green or red, A–J). Green or red fluorescence indicated isolectin-stained endothelial cells in blood vessels. On P12, the retinae were fully vascularized in both genotypes under room air (A, F). After OIR, a central avascular area was observed in both genotypes, indicating blood vessel regression (B–E, G–J). (M) Percentage of the central avascular area over the total retinal area in WT and AR^−/− OIR retinae at different postnatal days was estimated and compared. Significant reduction was observed in AR^−/− OIR retinae at all time points. In addition, significantly reduced central avascular area on P17 was observed in Fidarestat-treated retinae when compared with dH₂O-injected controls (K–L, N). n = 6 to 11 for each group. ***P < 0.001, **P < 0.01, unpaired t-test. Scale bar, 500 μm.

Neovascularization

The neovascular area in flat-mounted retinae (Figs. 5A, 5B) was measured, and intravitreal neovascular vessels on H&E-stained sagittal sections (arrows, Figs. 5C, 5D) were quantified. Significantly reduced neovascular area (Fig. 5E) and neovascular vessels (Fig. 5F) were observed in AR^−/− OIR retinae versus WT OIR retinae. In addition, a significantly reduced neovascular area was found in Fidarestat-injected retinae versus dH₂O-injected controls (Fig. 5G). 

Figure 5. 

 

Bucketed flat-mounted retinae (A, B) and H&E-stained retinal sagittal sections (C, D) for analyzing neovascularization after OIR on P17. Neovascular area was estimated in the flat-mounted retinae (A, B). Neovascular vessels were observed in both genotypes after OIR (arrows, C, D), and the number of neovascular vessels was quantified. Significantly reduced neovascular area (E) and neovascular vessels (F) were observed in AR^−/− OIR retinae when compared with WT OIR retinae. In addition, significantly reduced neovascular area on P17 was observed in Fidarestat-treated retinae when compared with dH₂O-injected controls (G). n = 7 to 10 for each group. *P < 0.05, unpaired t-test. Scale bar, 500 μm (B), 25 μm (D).

Retinal Blood Vessel Leakage

To investigate BRB breakdown, IgG extravasation was examined. IgG immunostaining detected outside the blood vessel lumen was regarded as vessel leakage. ²⁰ In normal WT and AR^−/− RA retinae, IgG staining was confined inside the blood vessel lumen, and no leaky vessels could be observed (Figs. 6A, 6C, 6E, 6G). After OIR, IgG staining was found outside the vessel lumen especially in WT OIR retinae (Figs. 6B, 6D, 6F, 6H). Significantly reduced leaky vessels in GCL and OPL were observed in AR^−/− OIR retinae versus WT OIR retinae (Figs. 6I, 6J). 

Figure 6. 

 

Representative images of IgG-stained retinal sagittal sections in GCL (A–D) and OPL (E–H) in WT and AR^−/− retinae on P17. In RA controls, the blood vessel lumen was intact and IgG staining was confined inside the lumen in both genotypes (A, C, E, G). However, after OIR, the vessel lumen was no longer intact and IgG staining was found outside the lumen, indicating blood vessel leakage (B, D, F, H). The number of leaky vessels was quantified and expressed as a percentage of total number of retinal vessels (I, J). Significantly reduced leaky vessels in GCL and OPL were found in AR^−/− OIR retinae when compared with WT OIR retinae. n = 5 to 8 for each group. **P < 0.01, *P < 0.05, unpaired t-test. Scale bar, 10 μm.

Oxidative Stress

NT Immunoreactivity.

NT was used as a marker for oxidative stress. Normally, NT immunoreactivity was minimal in the retinae of both genotypes (Figs. 7A, 7C, 7E, 7G). After OIR, NT immunoreactivity in GCL was increased (arrows, Figs. 7B, 7D, 7F, 7H) significantly for both genotypes (Fig. 7I). In addition, NT immunoreactivity was marked in INL of WT OIR retinae (arrowheads, Figs. 7B, 7D) and significantly reduced in AR^−/− OIR retinae versus WT OIR retinae (Fig. 7J). 

Figure 7. 

 

Immunohistochemical staining of nitrotyrosine (NT) antibody in WT (A–D) and AR^−/− (E–H) retinae on P17. Sections were counterstained with hematoxylin for nuclei. Minimal NT immunoreactivity was observed in RA controls (A, C, E, G). Increased NT immunoreactivity in GCL (arrows, B, D) and INL (arrowheads, B, D) was observed in WT OIR retinae. In AR^−/− OIR retinae, induced NT immunoreactivity was attenuated (F, H). (I) The IHC scoring system also demonstrated a significant increase in NT immunoreactivity in GCL after OIR for both genotypes. More importantly, significantly decreased NT immunoreactivity in INL was observed in AR^−/− OIR retinae when compared with WT OIR retinae (J). n = 5 to 8 for each group. **P < 0.01, *P < 0.05, Kruskal-Wallis test followed by Dunn's multiple comparison test. Scale bar, 25 μm.

PAR Immunoreactivity.

The level of PAR, the product of PAR polymerase (PARP) activated by DNA strand breaks, was also used as a marker for oxidative stress. ³⁷ Normally, PAR immunoreactivity was mainly present in the cytoplasm (Figs. 8A, 8C, 8E, 8G). In WT OIR retinae, PAR immunoreactivity was noted not only in the cytoplasm but also in the nuclei along the GCL in the whole retina (arrows, Figs. 8B, 8D). In AR^−/− OIR retinae, less PAR immunoreactivity was observed in the nuclei (Figs. 8F, 8H). The IHC score was given based on the intensity and location of PAR staining along the GCL in the whole retina. Significantly reduced PAR immunoreactivity was shown in AR^−/− OIR retinae versus WT OIR retinae (Fig. 8I). 

Figure 8. 

 

Immunohistochemical staining of PAR antibody in WT (A–D) and AR^−/− (E–H) retinae on P17 (A–H). Sections were counterstained with hematoxylin for nuclei. PAR immunoreactivity was observed in the cytoplasm in GCL in RA controls (A, C, E, G). After OIR, PAR immunoreactivity was also noted in the nuclei in GCL in both genotypes (arrows, B, D, F, H). (I) The IHC scoring system also showed significantly increased PAR immunoreactivity in WT OIR retinae. More importantly, a significant reduction was observed in AR^−/− OIR retinae when compared with WT OIR retinae. n = 7 or 8 for each group. *P < 0.05, Kruskal-Wallis test followed by Dunn's multiple comparison test. Scale bar, 25 μm.

Glial Responses

Normally, GFAP staining was localized in astrocytes around the blood vessels along the ILM (Figs. 9A, 9C, 9E, 9G). In WT OIR retinae, more intense GFAP immunoreactivity was observed in the ILM (Figs. 9B, 9D). The staining in astrocytes appeared to be expanded especially in peripheral retinae (Fig. 9D). In addition, strong GFAP immunoreactivity was observed in Müller cell processes across the IPL especially in the central retina (arrow, Fig. 9B). In AR^−/− OIR retinae, increased GFAP immunoreactivity was found in the ILM, and it appeared to be less expanded (Figs. 9F, 9H). In addition, less GFAP immunoreactivity in Müller cell processes was observed than that in WT OIR retinae (Figs. 9B, 9D, 9F, 9H). IHC scoring also demonstrated significantly increased GFAP immunoreactivity in WT OIR retinae and a trend toward increase in AR^−/− OIR retinae (Fig. 9I). Moreover, a trend toward decrease was shown in AR^−/− OIR retinae versus WT OIR retinae (Fig. 9I). 

Figure 9. 

 

Immunohistochemical staining of astrocytes and Müller cells with GFAP antibody in WT (A–D) and AR^−/− (E–H) retinae on P17. GFAP immunoreactivity was observed in astrocytes around the blood vessel lumen in RA controls (A, C, E, G). Increased GFAP immunoreactivity along ILM was observed in OIR groups when compared with RA controls for both genotypes (B, D, F, H). More importantly, intense GFAP immunoreactivity was present in Müller cell processes across central IPL in WT OIR retinae (arrow, B) but not in AR^−/− OIR retinae (F). (I) IHC scoring analysis also demonstrated significantly increased GFAP immunoreactivity in WT OIR retinae. Although no statistical difference was found in AR^−/− OIR retinae versus AR^−/− RA retinae, a trend toward increase was observed. In addition, a trend toward decrease was shown in AR^−/− OIR retinae versus WT OIR retinae. n = 5 to 8 for each group. ***P < 0.001, Kruskal-Wallis test followed by Dunn's multiple comparison test. Scale bar, 25 μm.

Inflammation

Microglial Response.

Immunohistochemical staining for Iba-1 antigen on P17 whole-mount retinae showed that ameboid- and ramified-form microglia were present throughout the retinae in both genotypes (Figs. 10A, 10B). Ameboid-form microglia showed large cell bodies, while ramified-form microglia showed small cell bodies with long processes. ³⁹ In addition, isolectin GS-IB4, which is also expressed as a surface marker by activated microglia, ^40,41 was costained with Iba-1. Normally, ramified-form microglia dominated while ameboid-form microglia were also observed occasionally (Fig. 10A). After OIR, the number of isolectin-immunopositive ameboid-form microglia was increased in the superficial layer of the central avascular area and midperipheral tuft area in both genotypes (Figs. 10B, 10C, 10D). Compared with what was observed in WT OIR retinae, the number of the ameboid-form microglia was significantly reduced in the central avascular area (Fig. 10C) but increased in the midperipheral tuft area (Fig. 10D) in AR^−/− OIR retinae. In the central retina, markedly increased ramified-form microglia, immunopositive for isolectin, were observed in deep layers in both genotypes but significantly reduced in AR^−/− OIR retinae versus WT OIR retinae (Figs. 10B, 10C). In the midperipheral nontuft area, increased ramified-form microglia, immunonegative for isolectin, were found in both genotypes (Figs. 10C, 10D). Moreover, in the deep layers of the midperipheral retina, significantly reduced ramified-form microglia were found in AR^−/− OIR retinae versus WT OIR retinae (Fig. 10D). 

Figure 10. 

 

Representative images of retinal areas stained with Iba-1 antibody and isolectin GS-IB4 in WT and AR^−/− retinae on P17. (A) Generally, ramified-form microglia dominated while occasionally ameboid-form microglia (arrows) were observed in WT and AR^−/− RA retinae. (B) After OIR, positive staining with both Iba-1 (red) and isolectin GS-IB4 (green) indicated activated microglia. In the central area, ameboid-form microglia (arrows, B) were observed in superficial layers while ramified-form microglia dominated in the deep layers. In the midperipheral nontuft area, mostly ramified-form microglia were observed in both layers. In the midperipheral tuft area, ameboid-form microglia were found in the superficial layers, and ramified-form microglia dominated in the deep layers. (C, D) The number of ameboid- and ramified-form microglia was quantified. Significant increase after OIR was observed in both genotypes. Compared with those in WT OIR retinae, ameboid-form microglia in AR^−/− OIR retinae were significantly reduced in the superficial layer of the central retinal area but increased in the superficial layer of the midperipheral tuft area. In addition, ramified-form microglia were significantly reduced in deep layers in AR^−/− OIR retinae compared with WT OIR retinae. (E) Western blot analysis demonstrated significantly induced p-IκB level in WT OIR retinae. More importantly, a significant reduction was found in AR^−/− OIR retinae when compared with WT OIR retinae. n = 4 to 8 for each group. ***P < 0.001, **P < 0.01, *P < 0.05, one-way ANOVA followed by Bonferroni's multiple comparison test. Scale bar, 50 μm.

p-IκB Expression.

In Western blot analysis, no significant difference was observed in expression of p-IκB, a pro-inflammatory response protein, between the two genotypes under room air (Fig. 10E). After OIR, p-IκB level was significantly increased in WT OIR retinae but not in AR^−/− OIR retinae. More importantly, p-IκB level was significantly smaller in AR^−/− OIR retinae versus WT OIR retinae (Fig. 10E). 

VEGF, p-Erk1/2, and p-Akt Expression

The expression of signaling molecules including VEGF, p-Erk1/2, and p-Akt involved in vascular endothelial cell proliferation and migration was examined (Figs. 11A, 11B, 11C). No significant difference in these three proteins was observed between the two genotypes under room air. Upon OIR, expression for these three proteins was significantly increased in both WT and AR^−/− retinae versus their respective RA controls (Fig. 11A). More importantly, significantly reduced induction of these three proteins was found in AR^−/− OIR retinae versus WT OIR retinae (Figs. 11A, 11B, 11C). 

Figure 11. 

 

Western blot analysis of VEGF, p-Erk1/2, and p-Akt expression in WT and AR^−/− retinae on P17 (A–C). No significant difference in all three protein levels was observed between WT and AR^−/− RA retinae (A–C). (A) After OIR, significantly increased VEGF expression was found in both genotypes. Significantly reduced VEGF level was observed in AR^−/− OIR retinae compared with WT OIR retinae. (B) Significantly increased p-Erk1/2 expression was shown in WT OIR retinae but not in AR^−/− OIR retinae. Significantly reduced p-Erk1/2 level was found in AR^−/− OIR retinae when compared with WT OIR retinae. (C) Significantly increased p-Akt expression was observed in WT OIR retinae. Significantly reduced p-Akt level in AR^−/− OIR retinae compared with WT OIR retinae was shown. n = 5 to 9 for each group. ***P < 0.001, **P < 0.01, *P < 0.05, one-way ANOVA followed by Bonferroni's multiple comparison test.

Discussion

In the present study, central obliteration, neovascularization, blood vessel leakage, and revascularization were observed in the mouse model of OIR as previously reported. ^4,7,35 More importantly, we provide strong evidence that AR deficiency was associated with reduced severity of these pathological features, suggesting a potential therapeutic role of AR inhibition in ischemia-induced retinopathy. 

It was recently reported that the rd8 mutation, a single nucleotide deletion in the Crb1 gene resulting in a form of retinal degeneration, ⁴² was identified in the C57BL/6N mouse substrain. ⁴³ Despite the fact that retinal degeneration in rd8 mice does not typically manifest in young pups, ⁴² the rd8 mutation is associated with retinal telangiectasia with exudation. ⁴⁴ As AR^−/− mice were maintained on the C57BL/6N background, these findings posed a major concern. A detailed analysis on the retinal structure was performed in neonates from various AR^−/− mouse litters, and no pathological features were observed. Comparable retinal vascular development and retinal structure were observed in WT and AR^−/− RA retinae from P7 to P17. There was no change in the thickness of various retinal layers, suggesting no retinal degeneration. There was also no vessel leakage in AR^−/− RA retinae on P17. Moreover, none of the cellular pathology reported by Mattapallil et al. was observed in our adult AR^−/− retina, whose morphology was previously published. ^29,43 Therefore, at this point we believe that the protective role of AR deficiency in the OIR model in our studies was not due to the rd8 mutation. Further thorough genotyping and sequencing through the Crb1 gene as has been previously reported would have to be performed. ⁴³  

It has been reported that the mouse model of oxygen-induced retinopathy has several known factors that can influence the outcome. ³⁴ In particular, the quality of maternal care during the course of an experiment and the resulting size and weight of the pups on P17 have a major effect on vascular obliteration and neovascularization. ³⁴ To reduce the runty phenotype, we limited the litter size to seven or eight pups for each mother as previously suggested by Connor et al. ³⁴ Moreover, pups from at least four litters were used to overcome biological variability. ³⁴ In our experiments, AR^−/− mice were observed to breed well, and the weight of pups on P7 (∼4 g) and P17 (∼6 g) was identical to that of WT mice with limited litter size. Therefore, the observed phenotypes reflected the pathological features observed in OIR as reported by others. ^4,7,35  

Previously, oxidative stress was highly implicated as playing a fundamental role in the pathogenesis of neovascularization. ^9,10,45 NT is the oxidative product of peroxynitrite with tyrosine residues of proteins; it is considered a useful marker for oxidative stress. ⁴⁶ In addition, increased reactive free radicals and peroxynitrite may cause DNA damage, leading to PARP activation. ²⁹ Increased NT accumulation and PARP activation after OIR were observed, providing further support for an association between oxidative stress and ischemia-induced retinal neovascularization. Previously, we showed that genetic deletion and the pharmacological inhibition of AR could reduce oxidative stress (evident by NT and PARP induction) in retinal ischemia/reperfusion. ²⁹ In the current study, we demonstrated similar findings in the mouse OIR model. More importantly, with AR deficiency, there was an observed reduction of oxidative stress that was associated with a reduction of central vaso-obliteration, neovascularization, and vascular leakage. 

Diabetes-induced sustained oxidative stress has been reported to be a major cause of retinal inflammation. ^15,19 In the animal models of OIR, induced inflammation has also been demonstrated. ^{16,41,47–49} Microglia, resident macrophages whose role is to maintain retinal integrity by constantly surveying the local environment and responding to injury or inflammation, ^41,47 have been shown to play a role in OIR. ^16,41,48,49 Upon activation, microglia undergo morphological change from ramified cells to ameboid phagocytes, ^41,47 accompanied by expression of several surface markers including those reactive to Griffonia simplicifolia isolectin B4. ^39,50 In addition, microglia secrete pro-angiogenic factors and cytokines. ⁵¹ Previous reports have demonstrated that microglia were present in avascular retina and their numbers were greatly7 elevated in the neovascular zone in OIR models. ^16,49 Our observations also showed increased activated microglia in the central avascular area. Compared with our observations in WT OIR retinae, revascularization occurred earlier in AR^−/− OIR retinae, and this early revascularization was associated with fewer microglia numbers observed in the central avascular area. These findings suggested that a certain degree of microglial activation might better facilitate revascularization, in line with the idea that a shift from a protective role induced by early microglial activation to harmful effects induced by overactivated microglia might occur. ³⁹  

In the superficial layer of the midperipheral tuft area, accumulation of activated microglia was observed, in line with previous publications. ^16,49 It was suggested that microglia played a phagocytic role to cause tuft regression. ^16,49 Our model showed that AR deficiency resulted in an increased microglial activation in association with a reduced tuft area. We postulate that the AR deficiency resulted in more activated microglia, thus directly causing increased tuft regression and thereby reducing neovascularization. 

Moreover, increased IκB phosphorylation further indicated the role of inflammation in ischemic retinopathy while AR deficiency attenuated IκB phosphorylation. Previously, pharmacological inhibition or genetic ablation of AR attenuated inflammatory signals in cell cultures or animal models of various inflammatory diseases. ^{15,19,31,32,52} Our observations indicated that the protective effects of AR deficiency on OIR were possibly mediated through modulation of the expression of IκB. 

Furthermore, glial responses have also been observed in the animal models of OIR. ^4,7,53,54 Normally, retinal glia including astrocytes and Müller cells provide the support and nutrients for retinal neurons, maintain BRB, and function as a template to guide the developing vasculature and neurons. ^41,50 In pathological circumstances, retinal glia became activated and resulted in “reactive gliosis.” ^40,55 GFAP upregulation is a well-known hallmark of reactive gliosis. Previous studies have shown an upregulated GFAP in glial cells in animal OIR models, ⁴ particularly in the avascular area. ⁵⁴ In addition, GFAP upregulation may lead to abnormal vascular growth. Previously, GFAP deficiency has been shown to specifically decrease newly formed neovascular blood vessels in the mouse model of OIR. ⁵³ In our study, the reduction in induced GFAP expression in AR^−/− OIR retinae indicated that AR deficiency may result in reduced neovascularization through downregulation of GFAP and reduction in Müller cell activities. 

In addition, Müller cell–derived VEGF has been reported to be a major contributor to ischemia-induced retinal neovascularization and vascular leakage. ⁷ Increased VEGF expression is a critical driving factor in neovascularization in ischemic ocular diseases. ^5,50,56 VEGF stimulation leads to Erk activation, which is important in endothelial cell proliferation, and inhibition of ERK has been shown to retard retinal neovascularization in ischemia retinopathy. ^57,58 In addition, VEGF stimulation induces Akt phosphorylation, which is important in endothelial cell migration. ⁵⁹ Inhibition of Akt phosphorylation effectively inhibited retinal endothelial cell survival and migration induced by VEGF in vitro ⁶⁰ and reduced neovascularization in vivo. ^61,62 Moreover, Erk and Akt activation were reported to be involved in VEGF-induced hyperpermeability. ⁶³ Previously, AR inhibition was shown to prevent VEGF-induced angiogenesis in vitro, ⁶⁴ suggesting a potential role of AR in angiogenesis. Our study demonstrated in an in vivo OIR model that AR deficiency led to reduction of VEGF, p-Erk1/2, and p-Akt expression that contributes directly to the reduction of neovascularization. 

Several AR inhibitors, such as Epalrestat, Ranirestat, and Fidarestat, have been tested in clinical trials in some diabetic complications. ⁶⁵ We chose Fidarestat for our present study because of its high potency and safety in the treatment of diabetic patients. ^66–68 In clinical settings, Fidarestat has not been tested for diabetic retinopathy and its possible inhibition of neovascularization. Our results suggest that there could be a possible role for AR inhibition in treating retinal neovascularization, via reduction of either oxidative stress or inflammation or both. 

Conclusion

The present study demonstrated that genetic deletion or pharmacological inhibition of AR reduced vascular damage in a murine model of oxygen-induced retinopathy. The protective role of AR deficiency was possibly a consequence of reducing oxidative stress, thereby modulating inflammatory responses and attenuating glial responses. These findings strongly suggested the therapeutic potential of AR inhibition in the treatment of ischemic retinopathy. 

Acknowledgments

The authors thank Chihiro Hibi at the Sanwa Kagaku Kenkyusho Co., Ltd., Japan, for the kind gift of Fidarestat. We also thank Antonia M. Joussen at the Department of Ophthalmology Charité, University Medicine Berlin, for her suggestions on the OIR model. 

Footnotes

 Supported by grants from the University Development Fund from The University of Hong Kong and Germany/Hong Kong Joint Research Scheme 2009/2010 (RGC Project No.: G_HK029/09).

Footnotes

 Disclosure: Z.J. Fu, None; S.-Y. Li, None; N. Kociok, None; D. Wong, None; S.K. Chung, None; A.C.Y. Lo, None